Introduction
The partnership between Cambridge Quantum Computing (now Quantinuum) and Roche pursued the application of quantum computing to pharmaceutical research. As one of the world’s largest pharmaceutical companies, Roche has consistently invested in cutting-edge technologies to enhance its drug discovery capabilities. The collaboration with CQC, a leading quantum software company specialising in quantum algorithms and applications, was established to explore how quantum computing could revolutionise the traditionally time-consuming and expensive process of drug development.
This partnership focused on developing quantum algorithms and software tools that could model molecular behaviour, predict drug-target interactions, and optimise lead compounds more accurately than existing classical methods. The collaboration brought together CQC’s expertise in quantum algorithms, particularly in quantum chemistry and machine learning, with Roche’s deep pharmaceutical knowledge and extensive drug development pipeline. This strategic alliance aimed to demonstrate practical quantum advantage in real-world pharmaceutical applications, potentially reducing the time from drug discovery to market by years and saving billions in development costs.
Challenge
The pharmaceutical industry faces enormous challenges in drug discovery and development, with the average cost of bringing a new drug to market exceeding $2.6 billion and taking 10-15 years. A significant portion of this time and expense is devoted to understanding molecular interactions, predicting drug efficacy, and identifying potential side effects. Classical computational methods struggle with the exponential complexity of modeling quantum mechanical systems, particularly when dealing with large biomolecules and their interactions with potential drug compounds. Roche specifically faced challenges in accurately simulating protein-drug interactions, optimizing molecular structures for desired properties, and predicting the behavior of drug candidates in biological systems. Traditional high-performance computing approaches were reaching their limits in handling the quantum nature of molecular interactions, leading to approximations that could miss crucial details. Additionally, the vast chemical space of potential drug compounds made exhaustive classical searching computationally prohibitive. The company needed more powerful computational tools to accelerate the identification of promising drug candidates, reduce the failure rate in clinical trials, and better understand the mechanisms of action for complex diseases. Quantum computing promised to address these challenges by naturally representing quantum mechanical systems and potentially providing exponential speedups for certain molecular simulation tasks.
Solution
CQC developed a comprehensive quantum software solution tailored to Roche’s drug discovery needs, leveraging their expertise in variational quantum algorithms and quantum machine learning. The solution centred around CQC’s proprietary EUMEN quantum chemistry platform and TKET software, which implemented advanced variational quantum eigensolver (VQE) algorithms optimised for near-term quantum devices. The team developed custom quantum circuits designed to simulate molecular Hamiltonians relevant to Roche’s drug targets, incorporating noise mitigation techniques to improve accuracy on current quantum hardware.
A key innovation was the development of hybrid classical-quantum algorithms that could efficiently handle the large molecular systems of pharmaceutical interest by identifying the most quantum-mechanically relevant portions for quantum simulation while treating the remainder classically. CQC also implemented quantum machine learning algorithms for molecular property prediction, using quantum kernel methods to identify patterns in molecular structure-activity relationships that classical methods might miss. The solution included a user-friendly software interface that allowed Roche’s computational chemists to specify molecular systems and run quantum simulations without needing deep expertise in quantum computing. Additionally, the platform incorporated advanced error mitigation strategies and resource estimation tools to maximise the utility of available quantum hardware while planning for future scaled-up implementations.
Implementation
The implementation of the quantum computing solution followed a phased approach, beginning with proof-of-concept studies on small molecular systems before scaling to pharmaceutically relevant targets. Initially, the teams focused on benchmarking quantum algorithms against classical methods for well-understood molecular systems, establishing baseline performance metrics and validating the quantum approach. CQC worked closely with Roche’s computational chemistry teams to identify specific use cases where quantum advantage would be most impactful, prioritizing protein-ligand binding affinity calculations and lead optimization problems. The implementation utilised a cloud-based quantum computing infrastructure, allowing Roche’s researchers to access various quantum hardware backends including IBM Quantum, Honeywell (now Quantinuum), and IonQ systems through CQC’s platform.
A crucial aspect of the implementation was the development of problem-specific decomposition strategies that mapped large pharmaceutical molecules onto available quantum hardware with limited qubit counts. The teams established workflows for data preprocessing, quantum circuit optimization, and results post-processing that integrated seamlessly with Roche’s existing computational infrastructure. Regular workshops and training sessions were conducted to upskill Roche’s computational teams in quantum computing concepts and the practical use of the developed tools. The implementation also included establishing metrics for comparing quantum and classical results, ensuring scientific rigour in evaluating the quantum advantage.
Results and Business Impact
The partnership yielded significant results in demonstrating quantum computing’s potential for drug discovery applications. Initial benchmarking studies showed that quantum algorithms could achieve chemical accuracy for small drug-like molecules using fewer computational resources than classical methods, particularly for systems with strong electron correlation effects. The quantum machine learning models developed showed improved prediction accuracy for molecular properties compared to classical baselines, with particularly notable improvements in predicting drug-protein binding affinities for certain target classes. While full quantum advantage for large pharmaceutical molecules remains contingent on improved quantum hardware, the hybrid algorithms developed provided immediate value by identifying quantum-critical regions of molecules for focused study.
The collaboration resulted in several peer-reviewed publications demonstrating quantum computing applications in drug discovery, establishing both companies as leaders in the field. From a business perspective, the partnership positioned Roche at the forefront of quantum computing adoption in pharmaceuticals, potentially providing a competitive advantage as quantum hardware continues to improve. The project also generated valuable intellectual property in quantum algorithms for drug discovery, with several patent applications filed jointly. The collaboration helped Roche build internal quantum computing expertise and establish a framework for evaluating and adopting quantum technologies as they mature. Preliminary estimates suggested that successful scaling of these quantum approaches could reduce drug discovery timelines by 20-30% for certain target classes.
Future Directions
Looking forward, the partnership outlined ambitious plans for scaling quantum computing applications as hardware capabilities improve. With the anticipated availability of fault-tolerant quantum computers with thousands of logical qubits within the next decade, the teams are developing algorithms capable of simulating entire protein-drug complexes without classical approximations. Future work includes expanding the quantum machine learning platform to incorporate more complex biological data, including genomic and proteomic information, to enable personalised medicine applications. The partnership is exploring quantum optimization algorithms for clinical trial design and patient stratification, areas where classical optimization faces significant challenges. As quantum hardware improves, the teams plan to tackle increasingly complex problems such as modeling drug metabolism, predicting drug-drug interactions, and simulating the effects of genetic variations on drug response. The collaboration is also investigating the potential of quantum computing for de novo drug design, using quantum generative models to explore chemical space more efficiently. Both companies are committed to contributing to the broader quantum computing ecosystem, with plans to open-source certain components of their quantum drug discovery platform to accelerate industry-wide adoption. The partnership serves as a model for academia-industry collaboration in quantum computing, with plans to expand to include additional pharmaceutical companies and quantum technology providers.